Bearing measurements on multi-moded signals using single snapshot data

Abstract

When measuring direction of arrival (DOA) in the presence of multi-path propagation, wavefronts corresponding to individual paths are often coherent. Direction finding algorithms which rely on the decomposition of the time averaged data covariance matrix are unable to resolve multiple sources without extra preprocessing, which reduces the array aperture. Therefore other direction finding tools are needed. Multidimensional methods capable of resolving signals in such a scenario, such as the deterministic maximum likelihood estimator (MLE) are computationally expensive to compute. This problem can be circumvented by employing a new algorithm, DOSE, (direction of arrival by signal elimination) which is both quicker than the MLE, and more robust than the decomposition methods. For uniform linear arrays it is shown that an even faster algorithm (which the authors call fast-DOSE) can be employed. Both of these algorithms process the 'single snapshot' data vector rather than the more usual procedure of employing the covariance matrix, thus achieving a significant computational advantage. A new, more robust method of calculating the relative frequencies between individual wavefronts is also presented. Experimental results using data from the HF scenario are presented to demonstrate the practical useability, accuracy and robustness of the new algorithms described.